One of the issues that marketing modelers often need to consider when fitting models to observational data is endogeneity, which is when there is unmodelled correlation between a model's error term and another variable or parameter in it. You can find a definition of endogeneity, and some possible causes of it, at:
https://en.wikipedia.org/wiki/Endogeneity
Provide examples of marketing analytics applications where endogeneity may be impacting modeling results, and explain what the likely cause(s) of endogeneity might be.
Endogeneity essentially occurs if the model has been underspecified. That is insufficient variables have been included in the model. In such a situation much of the variation remains unexplained and so that goes into the error term in such a situation if the left out variables are correlated with the variables already in the model then we have endogeneity bias and this leads to biased estimates. For instance if the sales of products are regressed on a marketing campaign undertaken and the salaries of individuals only per se, then this will mean that there will be a lot of the variation that will be left unexplained and so this will cause a correlation between these left out variables and those included and we have endogenity bias. This will cause biased estimates and also inflate standard errors and we will get insignificant results.
Get Answers For Free
Most questions answered within 1 hours.